Haha I'm glad. In my experience F1 itself is just as silly.. you can spend millions perfecting a car and simulating everything only for your driver to crash into a wall, which is exactly the kind of chaos that inspires this channel
Glad you introduced variance (or standard deviation) in the end. Saw that problem coming from quite early on, and then you mentioned Monza being mid after six categories.
@@MrVsGaragePerhaps doing a ranking for difference from 12.5 rather than an average would have made more sense. That way tracks with outliers would be punished because you definitely arent average if you max out one ranking and bottom out the other
@@slaterstimson yes, really thankful that a couple have reached out in the last few months. It's a big step towards the dream of making this channel a full time job!
Based on the information presented in the number of cornsers section, how about a video searching for the "Least Corneriest Corner" or the "Curviest Straight" That feels like the perfect pointless topic I'd love to watch a man explain his Excel sheet about for 30 minutes :)
As a former data analyst I would suggest to take the distance from 12# as the metric and not the actual placement from 1# to 24#. That way you don't count tracks which are extreme in both diterctions as avarage. For example, if a track ranks #2 in shortest and #22 in most turns per km, I wouldn't call it an average track, and the new metric would give a distance of 10 and 10, a opposed to a track who gets #13 and #8 for example, would get a distance score of 1 and 4, which more accurately represents what I think you meant as "averageness".
Best way I could think of would be to use the absolute values at both the 1st and 3rd quartile and give them a score of 100. Take the average of those two values and give it a score of 0. Now you have a two-way linear graph you can plot all the points into. This way you can account for very big gaps in absolute numbers on tracks that are only 1 position apart (looking at you monaco...). You'll end up with an averageness-score where lower number = more mid track.
For some reason, I feel like the only one who likes Barcelona as a track. That is from a driving point of view. I think sectors 1 and 2 are so fun to drive. And since they got rid of the chicane, sector 3 is also ok. Racing wasn't great of course though, I agree. But this and last year were good races with a lot of overtakes. To me the bad races of the past were more due to the cars, not due to the track.
As a kid I played Gran Turismo and the old F1 games a lot and Catalunya was always my favourite track to drive! But yeah it's just not suited for F1 cars, they're too big and fast for this track
I was preparing for the Excel madness where you would compare all the results of all the years of an f1 track against the final championship standings and see which ones the cars finish in the same order that they will finish in the championship.
the best way to define a straight in f1 is to assume, whether the given sector of the track can be approached with drs opened. You can't do that in Eau Rouge, but you certainly can do Baku last corners. stay with me now
Maybe its just video game nostalgia speaking, but I've always loved Cataluña precisely for its average-ness. It's a quintessential, downforce reliant racetrack, that in-game, is easy to understand and fun to drive. I don't mind that the racing isn't flashy, because I have an appreciation for the smoothness it takes to go fast, and for how impressive it is to pull those kinds of g's consistently, lap-in and lap-out.
genuinely the best f1 youtuber out there. 17 minutes felt like 5 its so engaging. i think this type of content is just much more fun than hearing the same news from a million different sources
14:45 made me really happy because this was my initial thought when you said you would use the average to determine 'averageness'. Another banger of a video!
Hi mate, excellent video! Is there any chance you would you share your excellent excel spreadsheet with all the data? I want to play about with the data and see if I find anything cool. I won't use it for evil, I pinky promise. Cheers!
Great video just one minor thing: if one the benefits of finding the most average track is to see if your car would perform well on it, what does attendance and ticket price have anything to do with that?
Great video just one minor thing: if one the benefits of finding the most average track is to see if your car would perform well on it, what does attendance and ticket price have anything to do with that?
YESSSSSSS THE MADNESS IS BACK LETS GOOOOOO, Always love a "we're gonna do all of them, baby" your efforts and dedication inspire me to try harder, Keep it up big guy ! Edit: I love the water splash transition, its toooo good
If you were to ever expand this idea here are some other categories: Wetness - % of wet sessions at the track Safety - Average number of safety cars/ virtual safety cars/ red flags that occur at the track each season Temperature - Average temperature of the track ( not the track surface) Consistency - how many times the track layout has changed (e.g a corner gets added/ removed between seasons) this should not include things like adding a new gravel trap or redesigning a barrier History - how long the track has been on the f1 calendar ( no. Of seasons) Difficulty - average no of drivers that dnf at the track each season, this should include dnfs due to mechanical failures as it shows the track is tough on the cars components ( this is quite similar to safty so you could just pick one or the other, or you could try find a way to combine them) Pitlane - how long the pit lane is in meters Run - distance between the start/ finish line and the first corner Competitiveness - how many overtakes occured the at the last race held there ( on include on track overtakes) Chonkyness - how wide the track is at its widest point ( or if it's easier how wide it is at the start/finish line Banking - the tracks steepest banked corner Cheekyness - no of track limit violations that occured at the tracks last held race
You should've tested the correlation of the average result for each driver in each circuit excluding the circuit being tested with the results on that circuit, and see which circuit was the most representative of results in all tracks.
I know this is late, but instead of averaging the rankings in these categories you should have given points for how far you were from the average. For example, the 12th ranking in a category should get 1st place points and the 24th ranking in a category should get 12th place points. You can choose the points system.
Montmeló is a great track: turn 1 to 3, turn 9 and 10, the last 2 turns since the Tilke's shame chicane was removed, the elevation changes. Its really nice stuff
I love your videos. You’re the only one that does these maths analyzations like you do. I’ve been subscribed from the beginning. Think you had maybe 500 at that point. Keep it up. You’re doing something different, which is important.
One of the most interesting ones in Saudi Arabia, being the Windiest track, but also having the 2nd highest average speed. Goes to show just how fast the corners are at that track.
Instead of summing and averaging their rank orders, could you try summing and averaging over how far their rank order is from 12.5? Otherwise a track that is #1 in half of the categories and #24 in the other half would unsatisfyingly (to me, at least) be deemed perfectly average, beating out a track that was #10 in every single category. sum(abs(12.5 - rank_in_category_i))/number_of_categories (Not sure how you're processing the data. This is just pseudocode. Also, I hope it doesn't get italicized by the underscores lol.) Edit: ah! You cover this at 14:33. I posted this about halfway through the video :-)
To be fair, one of those Monaco 2nd place wins is Verstappen starting from the front (but technically still grid position #2) in 2021. Which brings down the average to 1.67.
THANK YOU! I spent 10 minutes shouting VARIANCE at my screen while you walked through that dreadful methodology before you finally got to the part that made any statistical sense.
12:58 The 1964 Dutch Grand Prix had been won by Jim Clark, from p2 on the grid. Granted, you are probably only looking at the current track layout but still.
The reason Baku doesn't have the highest top speed is purely down to the setup of the cars. Because of sector 2, they can't use Monza-level front and rear wing planes. It's no question that they could be much faster than Monza by the time they reach turn 10, but they'd lose so much time in sector 2 that it's not worth it. Also adding that the main reason pole wins in Monaco less than half the last 10 years is because something happens to the polesitter that's out of their control. Lewis 2015, Danny 2016, Leclerc 2021 (though it was kind of his fault), Leclerc 2022. If it wasn't for that, the only time in the last 10 years the polesitter lost would have been 2017 when Vettel was just simply faster during the Ferrari pit window
Turbo engines lose very little performance due to elevation... Basicly, a turbocharger compensates for the lower air density, by simply compressing more air into the cylinder, losing only tiny fractions of overall performance. Plus, the lower air density is easier for cars to move through, so a turbocharged car like modern F1 cars, are actually faster on straights at high altitude, than the same car would be at sea level. It of course evens out a bit with the loss of aero that also becomes the result of lower air density: Slightly faster on straights, slightly slower in the corners... It was a huge problem in the WTCC for many years, where the NA cars were competetive or even dominant on sea level tracks, but as soon as you moved onto tracks at higher elevation, the turbo cars like the Hondas were fine, but the NA cars like the BMWs suddenly became complete jokes, as they could lose anywhere from 20 to nearly 100hp, just from track elevation and not having a turbo!
Typically, to find a prototypical datum in a set, you would calculate the centroid (average, dimension-wise) or medioid (median, dimension-wise) and then calculate the L2 distance (Euclidean distance) of each point to it. Basically, find the center (in some sense) of the data in a high-dimension space. Comparing ranks is *a thing*, but it doesn't fit in this case.
I was thinking how is he going to calculate the most average track? Is it going to be a complex piece of code that calculates averageness based on every track layout down to the length of every straight, angle of every corner, average speed at every part of the track... And then it was just a glorified tier list.
I did play around with Catalunya in RTB, taking into consideration how much land they have, the elevation changes, grandstands, access to grandstands, escape roads, marshalling posts etc etc. The change they made are pretty much the only kind that you can do on short notice and without drastically changing the whole layout of the track. I did check the changes Mr V did and i disagree with most of them (the 1st corner idea is good, the rest... difficult to do , and they pretty much destroy the classic track without offering any aid to the actual problems...), the actual changes are still better when we take ALL the stuff that tracks need to consider when making changes. It is easy to just draw lines on google maps, you need to look at everything else but the track to understand why.. the changes made were really the only changes they could've made. A huge problem is the last two corners, the wall on the outside is so close to the lot edge that there simply is no way to move then any further. And the last corner IS UNSAFE!.. but only a little bit, probabilities are quite low for a major catastrophic accident but the danger is there. Accidents in that corner are going to lead to red flags but so far, we haven't really had any, as it is quite easy corner. But... it is just a tiny bit too fast. It is not the only corner in the calendar that is borderline, Eau Rouge has too much compression and Radillion has unfavorable trajectory that should push walls much further out, but there is a sheer 40m drop on he other side... and it is classic. But, if we were allowed to do anything... Catalunya could have 30m hill!... It could have massive elevation change leading to a downhill braking and form a real backstraight that is close to main straight in length. You would go from 7th and cut thru everything between it and T12. Then do snake esses after it and we back to the old track at the last corner which would be taken at around 200kmh. It would also shorten the pitstop time quite a bit, giving it more tactical opportunities. I made a mockup and tried it in a sim, and it is a hoot but of course, cost about 100 million and couple of years.
Barcelona WAS pretty easily the most average prelayout engoodening. Much slower hairpin and the chicane really helped balance it out to that performance averaging aspect IMO. Also it's high deg but not bahrain, qatar, or hungary level
Austria already does that. I count eight, nine if they pay me. The second one is the one that is the first corner after T1, the one up the hill, not the slightly wonky straight. The next one after that is the one after the second straight. I could look over that straight and say the right hander after that is turn four (tho it isn't). What I don't understand is how they end up on 10. It's either 8 or 14. 10 means they count some wonky straights as corner while others that are just as much slightly not perfectly straight not. Im not upset or anything, but I really like to ask the person who came up with that number how and why. They either are very wise and can teach me their ways or utterly stupid and I can make them count things badly, for my entertainment. (Or they are French and wanted a more metric and decimal GP) I love your channel because it makes me look at the map of spielberg for 20 min, shaking my fist at it in frustration, while reflecting on the meaning of words like "ten" or "corner", "turn" even. You're like the magic mushrooms of F1 videos.
mr V posted this on the 28th of June i.e. the very beginning of the Austria race weekend. At the end of the video he questioned and predicted whether Merc and Ferrari could start challenging for wins too; since that day, every grand prix that has occurred (Austria and Britain) has been won by a Merc! funny how that worked out
@Mr V can I request a video?ive been lookinginto this myself but I am not able to find the data necessary and I do not have the time to look at every individual race report... But I am wondering if there is a way to look at race performance per driver in the rain. So I want to know who overachieves in the rain and who underachieves. We always hear about drivers like Verstappen, Hamilton, Button and even Stroll driving very well when it's wet. But are they actually better? And is it actually true that on a wet track "driver skill matters more"? Badically I want to know if drivers that are good in the wet actually do improve their average finisching position and by how much. Does full wet, intermediate or a drying track make a huge difference? And who is the absolute best and should hold the title of rain master? It can get pretty crazy once you try to figue out how to do this but step one is getting actual data on wet vs dry races and thats what ive not been able to find. So maybe you cangive it a try?
Honestly, a pretty interesting exercise. I wonder how many (sensible) means of manipulating the raw data gather end up with Spain as the least un-average track....
I think Azerbaijan counts as the most average track in F1 because the track has long straights,average windiness,medium speed corners,low speed corners and high speed corners.
Some of the weather aspects: - average temperature (preferably broken down in both track temperature and air temperature) - precipitation - wind speed Probably Barcelona would still be the most mid though after taking those into account
I think you could improve this by taking out the subjectiveness of ranking something as good (first) or bad (last), by making both first and last getting 1 point, 2nd and 23rd 2 points, 3rd and 22nd 3 point etc until 12th and 13th getting 12 points, and then the moist points being the most average
im suprised you didnt try to include corner characteristics as stats, % of high speed corners, or average corner speed, you covered straights but didnt cover the other bit of the tracks. tbf those sound like really hard statistics to get/calculate, would probably require raw telemetry data. G-forces would also be a wild stat (maximum and average) could be a great stat for driver strain
Love the video. This is the first video I have watched of yourself so hello! Thinking about the elevation stats it might be more indicative of performance to look at air density (air pressure and temperature ) rather than elevation as it is the reduction in air density that affects the power rather than elevation itself. But might be harder to find the stats as it's depended on weather on race weekend. Also might be thinking about temperature/track temperature for another stat. That would sit nicely into the model you have already implemented!
I feel like in the f1 games it’s instantly noticeable just how mid this track is and I actually like everything except the chicane but it’s just a very “familiar” track and the only memorable turns are 1/2 and the last 3 before the chicane. China is also there but it has the snail
You should definitely be summing up z-values and taking the lowest result. Variance is what you want to measure, because a track that scores extremely high in some categories and very low in others is going to look average.
I believe you should add one more category: temperature. I know it changes minute by minute but, you could find the average temp of the last ten years for example. Its an important factor because of tire degradation , engine overheating and driver exhaustion.
Question on the methodology for two of the categories. For pitlane loss, are you accounting for the different speed limits at various tracks? Which may or not be that relevant really since time lost is time lost. Secondly, elevation change should really be per kilometre, since it's about gradient not the total change.
Well, being the "averagest" circuit, and with 66 laps per race, it could be argued that it places the cars at their "real place" as it doesn't favor any kind of car desing particullary. And that that is it's appeal.
I personaly think that Imola is the most average. the times Mr V showed the whole graph, from 1 to 24, imola was never that extreme in either direction. Imola has low, medium and high speed and is pretty easy to get to.
Maaaaybe you could have deducted 12.5 points from each score and then calculated the absolute value. Because then you get how far away from the average score each track was for each stat, no matter if higher or lower. Then there's no need for variance and it would also be irrelevant if you ranked according to big numbers or low numbers for each stat and would get an undisputable result (probably Spain). Just an idea :) Good video nevertheless, I enjoy each one a lot!
You should merge all the average tracks together and make the most mid circuit of all time.
But what do we name such an average, uninspiring track?.. the Bo-ring?
@@MrVsGarage I was thinking of a better pun and I couldn't. The Bo-ring is much better than my "Mid-za"
The Average International Circuit, it's long but it surprisingly fits
@@adblox its called an oval
@@MrVsGarage Sounds kinda Swedish.
MEDIUM SPEED CORNERS
MEDIUM SPEED CORNERS
MEDIUM SPEED CORNERS
MEDIUM SPEED CORNERS
MEDIUM SPEED CORNERS
MEDIUM SPEED CORNERS
I love how your videos fill this odd void in me for statistic related f1 videos with little to no bearing on reality
Haha I'm glad. In my experience F1 itself is just as silly.. you can spend millions perfecting a car and simulating everything only for your driver to crash into a wall, which is exactly the kind of chaos that inspires this channel
Russell Singapore 2023😢@@MrVsGarage
Glad you introduced variance (or standard deviation) in the end. Saw that problem coming from quite early on, and then you mentioned Monza being mid after six categories.
I'm all for a silly unreliable result, but if I can do it properly and get an even sillier result then I absolutely will
@@MrVsGaragePerhaps doing a ranking for difference from 12.5 rather than an average would have made more sense. That way tracks with outliers would be punished because you definitely arent average if you max out one ranking and bottom out the other
@@DukeDonkey Did you watch the video? That's literally exactly what he did
thats what the variance Ranking does in the end If you Look at the blue Graph @@DukeDonkey
Where is Paul Ricard in all this? Boring as hell
I love how bro just apears out of nowhere out of the blue 💀
I'm like a chameleon I can turn my skin that colour whenever I want
@@MrVsGarage So would you say you're some kind of lizard person ? Danica was right all along...
@@Hawx32 we live in a simulation
Mr V has been cooking lately
Thank you! More good stuff coming! This was actually a last minute video as I had to push the next 2 back due to sponsors
@@MrVsGaragelove that you’re getting sponsors, is this a newer development?
@@slaterstimson yes, really thankful that a couple have reached out in the last few months. It's a big step towards the dream of making this channel a full time job!
17:16 man actually predicted everywthing
Based on the information presented in the number of cornsers section, how about a video searching for the "Least Corneriest Corner" or the "Curviest Straight"
That feels like the perfect pointless topic I'd love to watch a man explain his Excel sheet about for 30 minutes :)
You may (or may not) be surprised to learn that I have a long term project about something similar for a video...
@@MrVsGarageI feel like Austria turn 2 or Imola turn 1 would win this.
@@treytonlandplayz3151 Enter: The thingy before the pitlane in Jeddah
least corny corner
Baku and Monaco
When I’m in a never running out of ideas competition and my opponent is Mr V
is that why you just copy pasted a meme phrase from somewhere else.
@@TheEvilCheesecake yes.
He's like a pub chat god. But he would bring his spreadsheet to show his working
9:07 bro that was quite personal 🥺
As a former data analyst I would suggest to take the distance from 12# as the metric and not the actual placement from 1# to 24#.
That way you don't count tracks which are extreme in both diterctions as avarage.
For example, if a track ranks #2 in shortest and #22 in most turns per km, I wouldn't call it an average track, and the new metric would give a distance of 10 and 10, a opposed to a track who gets #13 and #8 for example, would get a distance score of 1 and 4, which more accurately represents what I think you meant as "averageness".
Would you rather take the distance in position or the difference to #12's value for that category?
Best way I could think of would be to use the absolute values at both the 1st and 3rd quartile and give them a score of 100. Take the average of those two values and give it a score of 0. Now you have a two-way linear graph you can plot all the points into. This way you can account for very big gaps in absolute numbers on tracks that are only 1 position apart (looking at you monaco...).
You'll end up with an averageness-score where lower number = more mid track.
Yep, take the centroid or medioid and find the nearest track to it in L2 terms
@@uvibalcey distance in position because you cannot easily normalize all metrics to same scale
@@nadavbentovim thank you!
"Medium Speed Corners" is this channel's iteration of SouthPawRacer's "2nd gear hairpin" 😅 Props to leaning into the "straight" joke as well!
But is this also Mr. v's most average video?
If this is 10/10 and all the rest are 10/10 then mathematically: yes
For some reason, I feel like the only one who likes Barcelona as a track. That is from a driving point of view. I think sectors 1 and 2 are so fun to drive. And since they got rid of the chicane, sector 3 is also ok. Racing wasn't great of course though, I agree. But this and last year were good races with a lot of overtakes. To me the bad races of the past were more due to the cars, not due to the track.
As a kid I played Gran Turismo and the old F1 games a lot and Catalunya was always my favourite track to drive! But yeah it's just not suited for F1 cars, they're too big and fast for this track
@@MrVsGaragei agree
Imagine just doing all those work just to end up in Spain without the s
We only eat Paella in this house because my whole life is (s)pain
9:10 hardest I've laughed in a long time Mr. V
reddit
So the most average F1 track was the friends we made along the way I guess ?
Please stop trying to drive a race car over my friends
@@MrVsGarage No 😎
How much do you love Microsoft Excel?
V: "yes"
new category: Ocon’s finishing position (per track)
Mr. V putting out bangers
I was put on this earth to drink milk and make bangers, and I'm all out of milk...
Makes sense
I was preparing for the Excel madness where you would compare all the results of all the years of an f1 track against the final championship standings and see which ones the cars finish in the same order that they will finish in the championship.
the best way to define a straight in f1 is to assume, whether the given sector of the track can be approached with drs opened. You can't do that in Eau Rouge, but you certainly can do Baku last corners. stay with me now
Maybe its just video game nostalgia speaking, but I've always loved Cataluña precisely for its average-ness. It's a quintessential, downforce reliant racetrack, that in-game, is easy to understand and fun to drive. I don't mind that the racing isn't flashy, because I have an appreciation for the smoothness it takes to go fast, and for how impressive it is to pull those kinds of g's consistently, lap-in and lap-out.
genuinely the best f1 youtuber out there.
17 minutes felt like 5 its so engaging.
i think this type of content is just much more fun than hearing the same news from a million different sources
14:45 made me really happy because this was my initial thought when you said you would use the average to determine 'averageness'. Another banger of a video!
WHAT ARE THE OTHERS WITH LOW VARIANCE CLOSE TO MID? We gotta know which tracks can be the new average
Hi mate, excellent video!
Is there any chance you would you share your excellent excel spreadsheet with all the data?
I want to play about with the data and see if I find anything cool.
I won't use it for evil, I pinky promise.
Cheers!
Hello! I share stuff through my discord discord.gg/Uvx2wtFxCF
but what is the other track that scores 13.1 too?
maybe silverstone
Great video just one minor thing: if one the benefits of finding the most average track is to see if your car would perform well on it, what does attendance and ticket price have anything to do with that?
Great video just one minor thing: if one the benefits of finding the most average track is to see if your car would perform well on it, what does attendance and ticket price have anything to do with that?
YESSSSSSS THE MADNESS IS BACK LETS GOOOOOO, Always love a "we're gonna do all of them, baby" your efforts and dedication inspire me to try harder, Keep it up big guy !
Edit: I love the water splash transition, its toooo good
If you were to ever expand this idea here are some other categories:
Wetness - % of wet sessions at the track
Safety - Average number of safety cars/ virtual safety cars/ red flags that occur at the track each season
Temperature - Average temperature of the track ( not the track surface)
Consistency - how many times the track layout has changed (e.g a corner gets added/ removed between seasons) this should not include things like adding a new gravel trap or redesigning a barrier
History - how long the track has been on the f1 calendar ( no. Of seasons)
Difficulty - average no of drivers that dnf at the track each season, this should include dnfs due to mechanical failures as it shows the track is tough on the cars components ( this is quite similar to safty so you could just pick one or the other, or you could try find a way to combine them)
Pitlane - how long the pit lane is in meters
Run - distance between the start/ finish line and the first corner
Competitiveness - how many overtakes occured the at the last race held there ( on include on track overtakes)
Chonkyness - how wide the track is at its widest point ( or if it's easier how wide it is at the start/finish line
Banking - the tracks steepest banked corner
Cheekyness - no of track limit violations that occured at the tracks last held race
Just gonna say your content is my absolute cup of tea.
Keep em coming Mr. V!
Only youtuber to not answer his own question and still be more entertaining than all of sky sports f1 team.
I feel like you need more views and subscribers. The current statistics seem off
You should've tested the correlation of the average result for each driver in each circuit excluding the circuit being tested with the results on that circuit, and see which circuit was the most representative of results in all tracks.
I know this is late, but instead of averaging the rankings in these categories you should have given points for how far you were from the average. For example, the 12th ranking in a category should get 1st place points and the 24th ranking in a category should get 12th place points. You can choose the points system.
You could give 12 points to p12 and p13, 11 to p11 and p14, 10 to p10 and p15... Until 1 point to p1 and p24
The most interesting F1 content on TH-cam
you need to start publishing your data, I'd love to see full rankings
Gonna be the proud owner of a Car Go Vroom Vroom cap soon woop woop! Really like that your merch is very neutral looking, it all looks brilliant!
Dude you’re way too smart to be making silly TH-cam videos. You need to be doing data analytics for a bank or something
Haha I actually did something similar for one of the largest car manufactures but they didn't want to pay me enough
If you can measure windiness, you can measure kinkiness. What's the kinkiest track on the calendar?
😏😏
Was expecting number of overtakes per race hehe
what would happen if F1 raced at Talledega
Never expected Ricky Gervais to cover F1
It's a shame you're not a nascar fan, i'd love to see all this besides the turns for it xD this was so fun!
Montmeló is a great track: turn 1 to 3, turn 9 and 10, the last 2 turns since the Tilke's shame chicane was removed, the elevation changes. Its really nice stuff
I love your videos. You’re the only one that does these maths analyzations like you do. I’ve been subscribed from the beginning. Think you had maybe 500 at that point. Keep it up. You’re doing something different, which is important.
Track above sealevel? Sad Zandvoort / Netherlands noises
One of the most interesting ones in Saudi Arabia, being the Windiest track, but also having the 2nd highest average speed. Goes to show just how fast the corners are at that track.
Instead of summing and averaging their rank orders, could you try summing and averaging over how far their rank order is from 12.5?
Otherwise a track that is #1 in half of the categories and #24 in the other half would unsatisfyingly (to me, at least) be deemed perfectly average, beating out a track that was #10 in every single category.
sum(abs(12.5 - rank_in_category_i))/number_of_categories
(Not sure how you're processing the data. This is just pseudocode. Also, I hope it doesn't get italicized by the underscores lol.)
Edit: ah! You cover this at 14:33. I posted this about halfway through the video :-)
Taking Eau Rouge flat out? Yeah ask Jacques Villeneuve and Ricardo Zonta how that went. Pretty awful might i add.
bro when spain appeared on the screen at 15:35 i heard a thunder right outside my house
To be fair, one of those Monaco 2nd place wins is Verstappen starting from the front (but technically still grid position #2) in 2021.
Which brings down the average to 1.67.
THANK YOU!
I spent 10 minutes shouting VARIANCE at my screen while you walked through that dreadful methodology before you finally got to the part that made any statistical sense.
my fav f1 statistics nerd 😎
12:58
The 1964 Dutch Grand Prix had been won by Jim Clark, from p2 on the grid.
Granted, you are probably only looking at the current track layout but still.
The reason Baku doesn't have the highest top speed is purely down to the setup of the cars. Because of sector 2, they can't use Monza-level front and rear wing planes. It's no question that they could be much faster than Monza by the time they reach turn 10, but they'd lose so much time in sector 2 that it's not worth it.
Also adding that the main reason pole wins in Monaco less than half the last 10 years is because something happens to the polesitter that's out of their control. Lewis 2015, Danny 2016, Leclerc 2021 (though it was kind of his fault), Leclerc 2022. If it wasn't for that, the only time in the last 10 years the polesitter lost would have been 2017 when Vettel was just simply faster during the Ferrari pit window
Turbo engines lose very little performance due to elevation... Basicly, a turbocharger compensates for the lower air density, by simply compressing more air into the cylinder, losing only tiny fractions of overall performance.
Plus, the lower air density is easier for cars to move through, so a turbocharged car like modern F1 cars, are actually faster on straights at high altitude, than the same car would be at sea level. It of course evens out a bit with the loss of aero that also becomes the result of lower air density: Slightly faster on straights, slightly slower in the corners...
It was a huge problem in the WTCC for many years, where the NA cars were competetive or even dominant on sea level tracks, but as soon as you moved onto tracks at higher elevation, the turbo cars like the Hondas were fine, but the NA cars like the BMWs suddenly became complete jokes, as they could lose anywhere from 20 to nearly 100hp, just from track elevation and not having a turbo!
Typically, to find a prototypical datum in a set, you would calculate the centroid (average, dimension-wise) or medioid (median, dimension-wise) and then calculate the L2 distance (Euclidean distance) of each point to it. Basically, find the center (in some sense) of the data in a high-dimension space. Comparing ranks is *a thing*, but it doesn't fit in this case.
Boy oh boy has this man’s copium worked. Though it was hilarious to hear “Verstappen won. Obviously.” That seems so far away now…
I was thinking how is he going to calculate the most average track? Is it going to be a complex piece of code that calculates averageness based on every track layout down to the length of every straight, angle of every corner, average speed at every part of the track... And then it was just a glorified tier list.
I did play around with Catalunya in RTB, taking into consideration how much land they have, the elevation changes, grandstands, access to grandstands, escape roads, marshalling posts etc etc. The change they made are pretty much the only kind that you can do on short notice and without drastically changing the whole layout of the track. I did check the changes Mr V did and i disagree with most of them (the 1st corner idea is good, the rest... difficult to do , and they pretty much destroy the classic track without offering any aid to the actual problems...), the actual changes are still better when we take ALL the stuff that tracks need to consider when making changes. It is easy to just draw lines on google maps, you need to look at everything else but the track to understand why.. the changes made were really the only changes they could've made. A huge problem is the last two corners, the wall on the outside is so close to the lot edge that there simply is no way to move then any further. And the last corner IS UNSAFE!.. but only a little bit, probabilities are quite low for a major catastrophic accident but the danger is there. Accidents in that corner are going to lead to red flags but so far, we haven't really had any, as it is quite easy corner. But... it is just a tiny bit too fast. It is not the only corner in the calendar that is borderline, Eau Rouge has too much compression and Radillion has unfavorable trajectory that should push walls much further out, but there is a sheer 40m drop on he other side... and it is classic.
But, if we were allowed to do anything... Catalunya could have 30m hill!... It could have massive elevation change leading to a downhill braking and form a real backstraight that is close to main straight in length. You would go from 7th and cut thru everything between it and T12. Then do snake esses after it and we back to the old track at the last corner which would be taken at around 200kmh. It would also shorten the pitstop time quite a bit, giving it more tactical opportunities. I made a mockup and tried it in a sim, and it is a hoot but of course, cost about 100 million and couple of years.
Barcelona WAS pretty easily the most average prelayout engoodening.
Much slower hairpin and the chicane really helped balance it out to that performance averaging aspect IMO. Also it's high deg but not bahrain, qatar, or hungary level
Austria already does that. I count eight, nine if they pay me. The second one is the one that is the first corner after T1, the one up the hill, not the slightly wonky straight. The next one after that is the one after the second straight. I could look over that straight and say the right hander after that is turn four (tho it isn't).
What I don't understand is how they end up on 10. It's either 8 or 14. 10 means they count some wonky straights as corner while others that are just as much slightly not perfectly straight not.
Im not upset or anything, but I really like to ask the person who came up with that number how and why. They either are very wise and can teach me their ways or utterly stupid and I can make them count things badly, for my entertainment. (Or they are French and wanted a more metric and decimal GP)
I love your channel because it makes me look at the map of spielberg for 20 min, shaking my fist at it in frustration, while reflecting on the meaning of words like "ten" or "corner", "turn" even. You're like the magic mushrooms of F1 videos.
I'd love to see this experiment but instead of using a rank for each variable and each track, you'd use a z-score of the value of each variable.
mr V posted this on the 28th of June i.e. the very beginning of the Austria race weekend. At the end of the video he questioned and predicted whether Merc and Ferrari could start challenging for wins too; since that day, every grand prix that has occurred (Austria and Britain) has been won by a Merc! funny how that worked out
Have you tried getting the "most average track" by comparing its GP results with final WDC standings for the year?
@Mr V can I request a video?ive been lookinginto this myself but I am not able to find the data necessary and I do not have the time to look at every individual race report... But I am wondering if there is a way to look at race performance per driver in the rain. So I want to know who overachieves in the rain and who underachieves.
We always hear about drivers like Verstappen, Hamilton, Button and even Stroll driving very well when it's wet. But are they actually better? And is it actually true that on a wet track "driver skill matters more"?
Badically I want to know if drivers that are good in the wet actually do improve their average finisching position and by how much. Does full wet, intermediate or a drying track make a huge difference? And who is the absolute best and should hold the title of rain master?
It can get pretty crazy once you try to figue out how to do this but step one is getting actual data on wet vs dry races and thats what ive not been able to find.
So maybe you cangive it a try?
Honestly, a pretty interesting exercise.
I wonder how many (sensible) means of manipulating the raw data gather end up with Spain as the least un-average track....
I think Azerbaijan counts as the most average track in F1 because the track has long straights,average windiness,medium speed corners,low speed corners and high speed corners.
Some of the weather aspects:
- average temperature (preferably broken down in both track temperature and air temperature)
- precipitation
- wind speed
Probably Barcelona would still be the most mid though after taking those into account
We will need an update for Las Vegas winning grid position after more races. It will be fun however if the average didn't chance much or at all.
I think you could improve this by taking out the subjectiveness of ranking something as good (first) or bad (last), by making both first and last getting 1 point, 2nd and 23rd 2 points, 3rd and 22nd 3 point etc until 12th and 13th getting 12 points, and then the moist points being the most average
im suprised you didnt try to include corner characteristics as stats, % of high speed corners, or average corner speed, you covered straights but didnt cover the other bit of the tracks.
tbf those sound like really hard statistics to get/calculate, would probably require raw telemetry data.
G-forces would also be a wild stat (maximum and average) could be a great stat for driver strain
Love the video. This is the first video I have watched of yourself so hello!
Thinking about the elevation stats it might be more indicative of performance to look at air density (air pressure and temperature ) rather than elevation as it is the reduction in air density that affects the power rather than elevation itself. But might be harder to find the stats as it's depended on weather on race weekend.
Also might be thinking about temperature/track temperature for another stat. That would sit nicely into the model you have already implemented!
Temp, track temp and likelihood of rain should all figure into the stat.
I feel like in the f1 games it’s instantly noticeable just how mid this track is and I actually like everything except the chicane but it’s just a very “familiar” track and the only memorable turns are 1/2 and the last 3 before the chicane. China is also there but it has the snail
You should definitely be summing up z-values and taking the lowest result. Variance is what you want to measure, because a track that scores extremely high in some categories and very low in others is going to look average.
I believe you should add one more category: temperature. I know it changes minute by minute but, you could find the average temp of the last ten years for example. Its an important factor because of tire degradation , engine overheating and driver exhaustion.
"How straight or not a track is" is basically the queerness of the track. (Yay pride month!)
No way I just went to the second most expensive f1 race on the calendar when I could’ve just gone to the third most expensive one a year earlier 😭
Question on the methodology for two of the categories. For pitlane loss, are you accounting for the different speed limits at various tracks? Which may or not be that relevant really since time lost is time lost. Secondly, elevation change should really be per kilometre, since it's about gradient not the total change.
Tracks really need to be ranked by their deviation from 12.5 for each category, not average of ranks.
E: nvm
Does anyone else agree that COTA would be a good testing track? I know seasons and winter and snow... but that would've been fun.
Well, being the "averagest" circuit, and with 66 laps per race, it could be argued that it places the cars at their "real place" as it doesn't favor any kind of car desing particullary. And that that is it's appeal.
Maybe using absolute distance to the mean would have solved the baku issue? Cool video regardless!
10:37
Every. 1.5. Oval. Track.
Vegas, Nashville, Etc
What was the track to the left of Baku on the variance chart? It said 10.9 on it's bar
I personaly think that Imola is the most average. the times Mr V showed the whole graph, from 1 to 24, imola was never that extreme in either direction. Imola has low, medium and high speed and is pretty easy to get to.
The “bosh” at 9:17 😭
Another way can be to count an average distance from 12.5th place in each category. Barcelona would have won anyway, of course
Such a shame chain bear doesn't make videosanymore because a Mr V Chainbear collab would be incredible
I fucking knew it would end up being Spain, it’s way too funny for it not to be the case
the copium went hard... just too hard and now Max decided to forget his age as he went back to 2019 Max
ngl I expected you to award points based on distance to middle - ie 12 and 13 getting like 12 points, 11 and 14 getting 11pts, etc. XD
Yaayyyyy I much prefer Melbourne as a season opener to Bahrain so that makes me happy :3
Why cant i pay with iDeal 😢😢😢😢 I want the Car go VROOM VROOM HATTTTTTTT
"have you considered that you might have a skill issue" bro roasted the f1 game community 😭😭
Maaaaybe you could have deducted 12.5 points from each score and then calculated the absolute value. Because then you get how far away from the average score each track was for each stat, no matter if higher or lower. Then there's no need for variance and it would also be irrelevant if you ranked according to big numbers or low numbers for each stat and would get an undisputable result (probably Spain). Just an idea :) Good video nevertheless, I enjoy each one a lot!
Just read a few more comments and saw that other people had the same idea :)